From 602ef2c8192255c65b2de9ac17f9a5d9489fa5e5 Mon Sep 17 00:00:00 2001
From: Gerhard Schlager <gerhard.schlager@discourse.org>
Date: Mon, 27 May 2024 22:33:30 +0200
Subject: [PATCH] FIX: Store special characters in permalink URL as encoded
 characters

see https://meta.discourse.org/t/permalink-not-working-with-cyrilics-symbols/301130
---
 app/models/permalink.rb       | 24 +++++++++++++++---------
 spec/models/permalink_spec.rb | 30 ++++++++++++++++++++++++++++--
 2 files changed, 43 insertions(+), 11 deletions(-)

diff --git a/app/models/permalink.rb b/app/models/permalink.rb
index 3758860372e..ca07c75be2c 100644
--- a/app/models/permalink.rb
+++ b/app/models/permalink.rb
@@ -7,7 +7,7 @@ class Permalink < ActiveRecord::Base
   belongs_to :tag
   belongs_to :user
 
-  before_validation :normalize_url
+  before_validation :normalize_url, :encode_url
 
   validates :url, uniqueness: true
 
@@ -66,10 +66,6 @@ class Permalink < ActiveRecord::Base
     find_by(url: normalize_url(url))
   end
 
-  def normalize_url
-    self.url = Permalink.normalize_url(url) if url
-  end
-
   def target_url
     return relative_external_url if external_url
     return post.relative_url if post
@@ -80,10 +76,6 @@ class Permalink < ActiveRecord::Base
     nil
   end
 
-  def relative_external_url
-    external_url.match?(%r{\A/[^/]}) ? "#{Discourse.base_path}#{external_url}" : external_url
-  end
-
   def self.filter_by(url = nil)
     permalinks =
       Permalink.includes(:topic, :post, :category, :tag, :user).order("permalinks.created_at desc")
@@ -92,6 +84,20 @@ class Permalink < ActiveRecord::Base
     permalinks.limit!(100)
     permalinks.to_a
   end
+
+  private
+
+  def normalize_url
+    self.url = Permalink.normalize_url(url) if url
+  end
+
+  def encode_url
+    self.url = UrlHelper.encode(url) if url
+  end
+
+  def relative_external_url
+    external_url.match?(%r{\A/[^/]}) ? "#{Discourse.base_path}#{external_url}" : external_url
+  end
 end
 
 # == Schema Information
diff --git a/spec/models/permalink_spec.rb b/spec/models/permalink_spec.rb
index 4aaa4d6cb8b..99f353a3896 100644
--- a/spec/models/permalink_spec.rb
+++ b/spec/models/permalink_spec.rb
@@ -13,14 +13,40 @@ RSpec.describe Permalink do
 
   describe "new record" do
     it "strips blanks" do
-      permalink = described_class.create(url: " my/old/url  ")
+      permalink = described_class.create!(url: " my/old/url  ")
       expect(permalink.url).to eq("my/old/url")
     end
 
     it "removes leading slash" do
-      permalink = described_class.create(url: "/my/old/url")
+      permalink = described_class.create!(url: "/my/old/url")
       expect(permalink.url).to eq("my/old/url")
     end
+
+    it "checks for unique URL" do
+      permalink = described_class.create(url: "/my/old/url")
+      expect(permalink.errors[:url]).to be_empty
+
+      permalink = described_class.create(url: "/my/old/url")
+      expect(permalink.errors[:url]).to be_present
+
+      permalink = described_class.create(url: "my/old/url")
+      expect(permalink.errors[:url]).to be_present
+    end
+
+    context "with special characters in URL" do
+      it "percent encodes any special character" do
+        permalink = described_class.create!(url: "/2022/10/03/привет-sam")
+        expect(permalink.url).to eq("2022/10/03/%D0%BF%D1%80%D0%B8%D0%B2%D0%B5%D1%82-sam")
+      end
+
+      it "checks for unique URL" do
+        permalink = described_class.create(url: "/2022/10/03/привет-sam")
+        expect(permalink.errors[:url]).to be_empty
+
+        permalink = described_class.create(url: "/2022/10/03/привет-sam")
+        expect(permalink.errors[:url]).to be_present
+      end
+    end
   end
 
   describe "target_url" do